The HRRR weather model

At a glance

What HRRR is

HRRR is built to answer a short question: what is going to happen in the next few hours? The grid spacing of 3 kilometers is fine enough that the model can draw individual thunderstorms, bands of heavy snow and the edge of a rain-snow line, instead of smoothing them into a blob.

It is also refreshed constantly. A new run begins every hour, using the latest observations, including radar data that are assimilated every 15 minutes. That hourly refresh is the reason HRRR is a favorite on days when the weather is changing quickly.

The trade-off is range. Most runs look ahead 18 hours, and four runs a day, at 00Z, 06Z, 12Z and 18Z, extend to 48 hours. After a day or so, the exact placement of individual storms drifts, and forecasters lean on ensembles and global models.

Where it does well

• It can show individual storms and bands. When you want to know whether a line of storms will reach your town this evening, or where a snow band sets up, a 3 km grid is the right size.
• It refreshes every hour, so it responds quickly to what is happening now.
• It offers fields built for storm days: simulated radar, helicity, updraft helicity, lightning density, echo tops, visibility and ceiling.
• Hourly timing helps you work out when something starts and stops.

Where it falls short

• Range. It stops at 18 hours on most runs and 48 hours on four runs a day. Past that, use another model.
• Details can drift. A model that draws individual storms can draw them in the wrong place. Treat a precise-looking storm as one possibility and look at several hourly runs.
• Coverage is limited to the United States and Alaska, so it will not help for the rest of the world.
• A single HRRR run is one run. Comparing consecutive runs shows whether it is steady or flipping.

What it is typically used for

• Timing the arrival and end of thunderstorms, rain or snow over the next several hours.
• Watching for snow bands and where they set up.
• Checking wind gusts, visibility and ceiling for the next day.
• Getting a forecast sounding for the next hours at a point.

How to see HRRR in 4070

HRRR is in Model Explorer under Mesoscale Models, with a run available for every hour. It is part of 4070 Pro, which starts with a seven-day free trial.

• Fields include 1-hour precipitation, rate of precipitation type, snow totals, 1 km and composite simulated radar, radar max, helicity, updraft helicity, 10 meter wind and gusts, surface and best CAPE and CIN, lightning density, echo tops, ceiling height, visibility and upper-air maps.
• Regions include the United States, Alaska and sectors such as the Northeast and the South Central U.S.
• Soundings: a supported HRRR map lets you click a point for a forecast skew-T, hodograph and plain-language read.
• Plumes use the seven most recent hourly HRRR runs at a point as members, so you can see how the model’s own forecast has been shifting.
• Radar: observed radar continues into HRRR forecast frames on the same timeline.

To read the sounding, see how to read a skew-T.

How it fits with the other models

HRRR is a short-range specialist. Use the GFS or ECMWF IFS for the pattern days ahead, and step down to HRRR when the storm is within a day or two. The NAM 3 km Nest is a second high-resolution model that runs a different schedule, and the NBM blends many models into one forecast. When the HRRR and the other high-resolution models agree on timing, that is more useful than any one of them alone.

Specifications change when agencies upgrade their models. The Sources below link to the producing agency. Official National Weather Service forecasts, watches and warnings remain the authority for what is expected where you live.